Impr oving the photostability of bright monomeric or … 2008 Nature...Impr oving the photostability of bright monomeric or ange and red ßuor escent pr oteins N athan C Shaner 1,5,
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Improving the photostability of bright monomericorange and red fluorescent proteinsNathan C Shaner1,5, Michael Z Lin1,2, Michael R McKeown1,2, Paul A Steinbach1,2, Kristin L Hazelwood4,Michael W Davidson4 & Roger Y Tsien1–3
All organic fluorophores undergo irreversible photobleachingduring prolonged illumination. Although fluorescent proteinstypically bleach at a substantially slower rate than manysmall-molecule dyes, in many cases the lack of sufficientphotostability remains an important limiting factor forexperiments requiring large numbers of images of single cells.Screening methods focusing solely on brightness or wavelengthare highly effective in optimizing both properties, but theabsence of selective pressure for photostability in such screensleads to unpredictable photobleaching behavior in the resultingfluorescent proteins. Here we describe an assay for screeninglibraries of fluorescent proteins for enhanced photostability.With this assay, we developed highly photostable variants ofmOrange (a wavelength-shifted monomeric derivative of DsRedfrom Discosoma sp.) and TagRFP (a monomeric derivative ofeqFP578 from Entacmaea quadricolor) that maintain most of thebeneficial qualities of the original proteins and perform asreliably as Aequorea victoria GFP derivatives in fusion constructs.
Substantial progress has recently been made in developing mono-meric or dimeric fluorescent proteins covering the visual spec-trum1–13, but although brightness and wavelength have beenprimary concerns, photostability has generally been an after-thought (with the notable exception of mTFP1; ref. 12). Conse-quently, many new fluorescent protein variants have relatively poorphotostability. The first-generation monomeric red fluorescentprotein, mRFP1 (ref. 1), although reasonably bright, was lessphotostable than its ancestor,Discosoma sp. DsRed14. In subsequentgenerations of mRFP1 variants (the ‘mFruits’), we observed seren-dipitous enhancement in photostability in some variants2, leadingus to believe that it would be possible to apply directed evolutionstrategies to this property as well.To extend the utility of fluorescent proteins, having optimized
them for many other properties, we developed a screening methodthat additionally assays photostability in a medium-throughputformat during directed evolution. Using a high-intensity lightsource, we photobleached entire 10-cm plates of bacteria expressingthe fluorescent proteins of interest and selected those that main-
tained the most brightness. This approach allowed us to screenlibraries containing up to 100,000 clones reliably with no observedfalse-positive hits and to select simultaneously for the mostphotostable mutants that also maintained an acceptable level offluorescence emission at the desired wavelength, minimizing thetradeoff of desirable properties that frequently results from single-parameter screens. We applied our photostability screening assay tothe directed evolution of variants derived from the bright redmonomeric red fluorescent protein TagRFP and the fast-bleachingmonomeric orange fluorescent protein mOrange. The resultingvariants, TagRFP-TandmOrange2, were ninefold and 25-foldmorephotostable than their respective ancestors, and both made excel-lent fusion partners when expressed in mammalian cells.
RESULTSPhotostability assay and rationaleTo photobleach large numbers of bacterial colonies, we used a solarsimulator, which produces a collimated beam approximately 10 cmin diameter with light intensities of 95 or 141 mW/cm2 with525–555 (540/30) or 548–588 (568/40) nm bandpass filters, respec-tively. This intensity, although approximately 100-fold lower thanthat produced by unattenuated arc lamp illumination and 105-foldlower than instantaneous intensities during confocal laser illumi-nation, was sufficient to photobleach the photolabile fluorescentprotein mOrange to 50% initial intensity after approximately10 min. This reasonably short time allowed us to quickly screenbacterial libraries of up to 100,000 clones on plates. We minimizedthe heating of plates by placing them on a custom-built water-cooled aluminum block. At wavelengths necessary to photobleachorange and red fluorescent proteins, we found no substantialdecrease in bacterial viability after 2 h of illumination.
Evolution of a brighter photostable red monomerTo create a better red monomer, we initially undertook a rationaldesign approach, drawing on analysis of mCherry’s enhancedphotostability and mOrange’s higher quantum yield relative tomRFP1. Six generations of directed evolution with constantphotostability selection yielded the variant ‘mApple’, which, though
RECEIVED 21 NOVEMBER 2007; ACCEPTED 1 APRIL 2008; PUBLISHED ONLINE 4 MAY 2008; DOI:10.1038/NMETH.1209
1Department of Pharmacology, 2Howard Hughes Medical Institute and 3Department of Chemistry and Biochemistry, University of California at San Diego, 9500 GilmanDrive, La Jolla, California 92093, USA. 4National High Magnetic Field Laboratory and Department of Biological Science, The Florida State University, 1800 East PaulDirac Drive, Tallahassee, Florida 32310, USA. 5Present address: The Salk Institute for Biological Studies, 10010 N. Torrey Pines Rd., La Jolla, California 92037, USA.Correspondence should be addressed to R.Y.T. ([email protected]).
substantially brighter than mCherry, displayed complex photo-switching behavior (Fig. 1 and Tables 1 and 2, and SupplementaryFig. 1 and Supplementary Note 1 online). This behavior was morepronounced with continuous wide-field than with laser-scanningillumination and could be largely eliminated by excitation atalternate wavelengths or by intermittent illumination. However,given our later results using the brighter TagRFP as startingmaterial, we chose not to pursue mApple any further.Although the recently developed orange-redmonomer TagRFP13
exhibits remarkable brightness, we have found that its photostabil-ity is still far from optimal. In both our standard arc-lampphotobleaching and laser-scanning confocal assays, we determinedthat TagRFP bleaches approximately threefold faster than mCherry(Fig. 1a,b and Table 1). Thus, we chose this protein as anotherstarting point for improvement of photostability. We firstattempted rational design of a mutant library guided by the crystal
structure of the closely-related protein eqFP611 (ref. 15). With therationale that chromophore-interacting residues could influencephotostability, we performed saturation mutagenesis of Ser158 andLeu199, two residues proximal to the TagRFP chromophore. Wethen screened this library in bacteria with our solar simulator–based assay, using the 540/30 nm bandpass filter and exposuretimes of 120 min per plate, imaging the plates before andafter bleaching to select those colonies that displayed highabsolute brightness and a high ratio of post-bleach to pre-bleachfluorescence emission.From this directed library, we identified one clone, TagRFP
S158T (designated TagRFP-T), which had a photobleaching half-time of 337 s by our standard assay, making it approximatelyninefold more photostable than TagRFP (Fig. 1a–c and Table 1).TagRFP-T, which we further modified by appending GFP-like Nand C termini, possesses identical excitation and emission
0
250
500
750
1,000
1,250
0 100 200 300 400 500
Em
issi
on(p
hoto
ns/(
s !
mol
ecul
e))
Time (s)
0
250
500
750
1,000
0 250 500 750 1,000
Em
issi
on(p
hoto
ns/(
s !
mol
ecul
e))
Time (s)
0
250
500
750
1,000
0 100 200 300 400 500 600 700 800 900 1,000
Em
issi
on (
phot
ons/
(s !
mol
ecul
e))
Time (s)
-
mRFP1 Normoxic mOrange
EGFP/mEGFP
mCherry
tdTomato
mOrange
mKO
TagRFP
mApple
mOrange2
TagRFP-T
O2-free mOrange2
Normoxic mOrange2O2-free mOrange
Normoxic TagRFP
O2-free TagRFP-T
Normoxic TagRFP-TO2-free TagRFP
0
250
500
750
1,000
1,250
0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000
Em
issi
on (
phot
ons
/(s
! m
olec
ule)
)
Time (s)
a b c
d
Figure 1 | Comparison of photobleaching curves. (a) Arc-lamp photobleaching curves for mRFP1, EGFP, mCherry,tdTomato, mOrange, mKO, TagRFP, mApple, mOrange2 and TagRFP-T, as measured for purified protein and plotted asintensity versus normalized total exposure time with an initial emission rate of 1,000 photons/s per molecule.(b) Normalized laser scanning confocal microscopy bleaching curves for the same proteins (except for EGFP, which inthis case is the monomeric A206K variant) fused to histone H2B and imaged in live cells. The time axis representsnormalized total imaging time for an initial scan-averaged emission rate of 1,000 photons/s per molecule.(c,d) Arc-lamp photobleaching curves for TagRFP and TagRFP-T (c), and for mOrange and mOrange2 (d) undernormoxic and O2-free conditions (normalized as in a).
Table 1 | Physical and optical properties of new photostable fluorescent protein variants
TagRFP-T 555 584 81,000 0.41 33 4.6 100 min 337 44600 6,900aBrightness of fully mature protein, (extinction coefficient ! quantum yield)/1,000. bTime to bleach to 50% emission intensity under arc-lamp illumination, at an illumination level that causes each molecule toemit 1,000 photons/s initially, as measured in our lab. See reference 16 for details. cWith arc lamp illumination, equilibrated under O2-free conditions.
dTime to bleach to 50% emission intensity measured duringlaser scanning confocal microscopy, at an average illumination level over the scanned area that causes each molecule to emit an average 1,000 photons/s initially, as measured in our lab. A 543-nm laser linewas used for all proteins except mEGFP, which was bleached with a 488-nm laser (see Supplementary Methods for detailed description of normalization). eND, not determined. fAll measurements were performedin our lab.
wavelengths, quantum yield and maturation time to TagRFP, withonly a slightly lower extinction coefficient (81,000 versus98,000 M–1 cm–1) and a higher fluorescence pKa, the pH valueat which the fluorescent protein exhibits half-maximalfluorescence emission (4.6 versus 3.1). We expect that the benefitof increased photostability should offset the small decrease inbrightness and increase in acid sensitivity in most applications.Additionally, TagRFP-T matures to apparent completion andhas virtually no emission in the green region of the spectrum(Supplementary Fig. 1), making it suitable for co-imagingwith green fluorescent proteins. We verified that TagRFP-Tremainsmonomeric by gel filtration (data not shown). Because theS158T mutation is in the interior of the folded protein, weanticipated that TagRFP-T would perform nearly identically toTagRFP when used as a fusion tag. Indeed, live-cell imagingconfirmed that TagRFP-T does not interfere with localization ofany fusions tested (Fig. 2).Photobleaching of TagRFP and TagRFP-T under oxygen-free
conditions revealed that TagRFP-T’s photobleaching remainsoxygen-sensitive (Fig. 1c and Table 1). However, the oxygen-freebleaching half-time for TagRFP is similar to the ambient oxygen
bleaching half-time for TagRFP-T. We next compared TagRFP andTagRFP-Tas fusions to histone H2B expressed in living cells underconfocal illumination (Fig. 1b and Table 1). TagRFP-T had aphotobleaching half-time approximately ninefold greater thanthat of TagRFP, consistent with the results obtained for purifiedproteins under continuous wide-field illumination.
Evolution of a photostable orange monomerWe next attempted to engineer a photostable variant of mOrange,which is the brightest of the previously engineered mRFP1 variantsbut exhibits relatively fast bleaching. Because substitutions atposition 163 improved photostability during the evolution ofmCherry and mApple, we initially tested the M163Q mutant ofmOrange, but found that improved photostability was accompa-nied by undesirable decreases in quantum yield and maturationefficiency. The M163K mutant of mOrange exhibited enhancedphotostability and matured very efficiently, but suffered fromincreased acid sensitivity (pKa of B7.5). Because another orangefluorescent protein, mKO (derived from Fungia concinna)6, is bothhighly photostable16 and possesses a methionine at the positionequivalent to 163, we reasoned that other pathways must exist forincreasing photostability.To explore alternative photostability-enhancement evolution
pathways, we used iterative random and directed mutagenesisand selection using the solar simulator. Initially we screened arandomlymutagenized library ofmOrange by photobleaching with540/30 nm light for 15–20 min per plate (a time sufficient to bleachmOrange to B25% of its initial brightness) and selecting thebrightest post-bleach clones by eye. This screen identified a single
a b c d
e f g h
i j k l
m n o p
Figure 2 | Fluorescence imaging of TagRFP-Tsubcellular targeting fusions. (a–g) N-terminalfusion constructs (linker amino acid lengthindicated by the numbers): TagRFP-T–N1(a; N-terminal fusion cloning vector; expressionin nucleus and cytoplasm with no specificlocalization); TagRFP-T–7–cytochrome c oxidase(b; mitochondria human cytochrome c oxidasesubunit VIII); TagRFP-T–6–histone H2B(c; human; showing two interphase nuclei and onenucleus in early anaphase); TagRFP-T–7–b-1,4-galactosyltransferase (d; golgi; N-terminal81 amino acids of human b-1,4-galactosyltransferase); TagRFP-T–7-vimentin(e; human); TagRFP-T–7-Cx43 (f; rat a-1connexin-43); and TagRFP-T–7-zyxin (g; human).(h–p) C-terminal fusion constructs (linker aminoacid length indicated by the numbers): annexin(A4)–12–TagRFP-T (h; human; illustrated withionomycin-induced translocation to the plasmaand nuclear membranes); lamin B1–10–TagRFP-T(i; human); vinculin-23–TagRFP-T (j; human);clathrin light chain–15–TagRFP-T (k; human);b-actin–7–TagRFP-T (l; human); PTS1-2–TagRFP-T(m; peroximal targeting signal 1); RhoB-15–TagRFP-T (n; human RhoB GTPase with an N-terminal c-Myc epitope tag; endosome targeting);farnesyl-5–TagRFP-T (o; 20-amino-acidfarnesylation signal from c-Ha-Ras); andb-tubulin–6–TagRFP-T (p; human). All TagRFP-Tfusion vectors were expressed in HeLa (CCL-2)cells. Scale bars, 10 mm.
Table 2 | Mutations of new photostable fluorescent protein variants
clone, mOrange F99Y, which had approximately twofoldimproved photostability (data not shown). Saturation mutagenesisof residue 99—and residues 97 and 163, which we imagined couldhave synergistic interactions with residue 99—did not yieldadditional improvements.We then constructed a randomly mutagenized library of mOr-
ange F99Yand screened with a longer illumination time of 40 minper plate. This round of screening identified an additional muta-tion, Q64H, which conferred about a tenfold increase in photo-stability over the mOrange F99Y single mutant. Again, saturationmutagenesis of residues 64 and 99 along with neighboring residues97 and 163 did not produce clones that were improved over theoriginal clone identified in the random screen. Additionally,we found that the Q64H mutation alone did not confer substan-tially enhanced photostability but required the presence of theF99Y mutation (data not shown). Two additional rounds ofdirected evolution with continued selection for photostability(540/30 nm filter, 40 min per plate) improved the folding efficiencywith mutations E160K and G196D, giving the final clone,mOrange2 (Table 2).The highly desirable increase in photostability achieved in
mOrange2 is balanced by a modest decrease in quantum yield(0.60 versus 0.69) and extinction coefficient (58,000 versus 72,000M–1 cm–1), together corresponding to a30% decrease in brightness compared tomOrange. It also exhibits slightly shiftedexcitation and emission peaks (549 nmand 565 nm) and an increased maturation
half-time (4.5 h versus 2.5 h; Table 1). However, its photostabilityunder arc-lamp illumination is over 25-fold greater than that ofmOrange (Fig. 1d), making it nearly twice as photostable asmKO6, the previously most photostable known orange mono-mer16, approximately sixfold more photostable than TagRFP13
and about 1.3-fold more photostable than enhanced GFP(EGFP)16 (Fig. 1 and Table 2). During laser-scanning confocalimaging, mOrange2 was approximately sixfold more photostablethan mOrange and threefold more photostable than mKO(Fig. 1b). Notably, the brightness and maturation time of mOr-ange2 are quite similar to those for mKO. mOrange2 remainsacid-sensitive with a pKa of 6.5, making it undesirable for targetingto acidic compartments, but attractive as a possible marker forexocytosis or other pH-variable processes17. Also, because itcontains a small fraction of immature (but nonfluorescent) chro-mophore (Supplementary Fig. 1), mOrange2 may not be anideal FRETacceptor. As with TagRFP-T, we verified that mOrange2remained monomeric using gel filtration (data not shown).We then investigated the role of the key photostability-enhancingmutations present in mOrange2, tested it in a wide rangeof fusion constructs, and compared its performance withthat of mKO and tdTomato (Fig. 3 and SupplementaryNote 2 online).
a b c d
e f g h
i j k l
m n o p
q r s t u
Figure 3 | Widefield fluorescence imaging ofmOrange2 subcellular targeting fusions. (a–h) N-terminal fusion constructs (linker amino acidlength indicated by the numbers): mOrange2-17-keratin (a; human cytokeratin 18); mOrange2-7-Cx26 (b; rat b-2 connexin-26); mOrange2-7–b-1,4-galactosyltransferase (c; golgi; N-terminal81 amino acids of human b-1,4-galactosyltransferase); mOrange2-7-vimentin(d; human); mOrange2-7-EB3 (e; humanmicrotubule-associated protein; RP/EB family);mOrange2-7–cytochrome c oxidase(f; mitochondria; human cytochrome c oxidasesubunit VIII); mOrange2-22-paxillin (g; chicken);and mOrange2-19–a-actinin (h; human non-muscle). (i–p) C-terminal fusion constructs (linkeramino acid length indicated by the numbers):lamin B1–10-mOrange2 (i; human); b-actin–7-mOrange2 (j; human); glycoprotein 1–20-mOrange2 (k; rat lysosomal membraneglycoprotein 1); peroxisomal targeting signal 1–2-mOrange2 (l); b-tubulin–6-mOrange2 (m; human);fibrillarin-7-mOrange2 (n; human); vinculin-23-mOrange2 (o; human); and clathrin light chain-15-mOrange2 (p; human). (q–u) Laser scanningconfocal images of HeLa cells expressing histoneH2B–6-mOrange2 (N-terminal fusion; human)progressing through interphase (q), prophase (r),prometaphase (s), metaphase (t) and earlyanaphase (u). The cell line used for expressingmOrange2 fusion vectors was Gray fox lungfibroblast cells (FoLu) in e and j, and humancervical adenocarcinoma cells (HeLa) in theremaining panels. Scale bars, 10 mm.
Evaluation of reversible photoswitchingBecause of concerns that our screening method might select forphotoswitching behavior, we tested our selected variants as well asother commonly used fluorescent proteins using both widefieldand confocal imaging. Nearly all had some degree of reversiblephotoswitching, which we observed as a recovery of up to 100% ofpre-bleach fluorescence intensity when the fluorescent protein wasbleached to B50% of its initial intensity and then observed againafter 1–2 min without illumination. In fact, several commonly usedA. victoria GFP variants including EGFP, Cerulean and Venus,displayed reversible photoswitching18 more severe than thatobserved for the variants we identified. A summary table of theresults of these experiments along with representative traces forTagRFP, TagRFP-T, EGFP and Cerulean are available in Supple-mentary Note 3 online. These results suggest that our screen is notselecting specifically for photoswitching, which is no worse inthe new proteins (except for mApple) than in well-establishedfluorescent proteins.Although our observation of reversible photoswitching in such a
broad range of fluorescent proteins certainly raises concerns aboutthe potential for previously undetected experimental artifacts, it isbeyond the scope of this study to determine how common or severethis phenomenon may be. Of particular concern is the implicationthat fluorescence recovery after photobleaching experiments maybe prone to artifacts that would confound data interpretation. Weperformed a limited evaluation of this possibility using histoneH2B fusions to EGFP and EYFP expressed in mammalian cells andimaged on a laser-scanning confocal microscope. When webleached these proteins to near completion with full laserpower and then observed for recovery, we observed a negligibleamount of reversible photoswitching (data not shown). How-ever, an in-depth investigation is warranted to rule out such aneffect in other fluorescent proteins and under more variedexperimental conditions.
DISCUSSIONAlthough the precise kinetics of photobleaching for a givenfluorescent protein are strongly dependent on illumination inten-sity and temporal regimen, we found that improvements inphotostability at B0.1 W/cm2 usually qualitatively predictimproved performance under typical conditions for wide-fieldand laser scanning microscopy. The exceptions were mApple’sreversible photoswitching (Supplementary Note 1) and tdToma-to’s poor performance under laser scanning confocal illumination(Fig. 1b). Also, our screen used bacteria to express fluorescentprotein libraries, but all proteins produced from these studiesbehaved similarly when later tested in purified form or expressedin mammalian cells, consistent with our previous experience.Fluorescent proteins had been photobleached using an array of
LEDs during the evolution of mTFP1 to select against unacceptablephotolability or photoswitching, resulting in a protein with ableaching half-time 110 s12. We applied photostability as a primarycriterion to improve multiple fluorescent proteins, and our resultsdemonstrate that high photostability is a selectable phenotype.Moreover, a solar simulator takes advantage of the strong mercurylines at 546, 577 and 579 nm and allows greater flexibility in thechoice of excitation wavelength than would be possible with LEDs.Although it is difficult to draw strong conclusions about exact
mechanisms of photobleaching from the mutations that confer
photostability to mOrange2, specific regions proximal to thechromophore appear to influence the modes of photobleachingthe protein is able to undergo. DsRed, when illuminated by a532-nm pulsed laser, undergoes decarboxylation of Glu215, as wellas cis-to-trans isomerization of the chromophore19. Such chromo-phore isomerization has been implicated in the photoswitchingbehavior of Kindling fluorescent protein (KFP)20,21 and Dronpa5,22
as well as predecessors to mTFP1 (refs. 12 and 23). Decarboxylationof the corresponding glutamate (position 222) in A. victoria GFPalso leads to changes in optical properties24–26. However, ourobservation that oxidation is important in mOrange, TagRFP andTagRFP-T photobleaching suggests that chromophore isomeriza-tion and Glu215 decarboxylation may have only a minor role forsuch proteins under normoxic conditions. Additionally, we foundno evidence by mass spectrometry that photobleaching using thesolar simulator led to any detectable decarboxylation of Glu215 inmOrange (data not shown). Under some conditions mOrange2shows an initial photoactivation of about 5% (Fig. 1a,d) beforebleaching takes over. At present we have no molecular explanationfor this effect or the reversible photoswitching that is common tomost fluorescent proteins (Supplementary Note 3).For mRFP1 variants, we observed the importance of residue 163
in influencing photostability (Supplementary Note 1) but alsoobserved somewhat context-specific effects of residue 163 andsurrounding residues on different wavelength-shifted variants.This region, composed of residues 64, 97, 99 and 163, appears tobe important in determining photostability. However, of these, onlyresidue 163 is in direct contact with the chromophore. It may bethat the mutations Q64H and F99Y together lead to a rearrange-ment of the other side chains in the vicinity of the chromophore soas to hinder a critical oxidation that leads to loss of fluorescence.Discrepancies in tubulin and connexin localization (Supplemen-
tary Note 2) when fused to mOrange2 versus mKO or tdTomatocan probably be attributed to the three-dimensional structure of thefluorescent protein and potential steric hindrance in the fusions.mOrange2 contains extended N and C termini derived from EGFPto improve performance in fusions, whereas the much shorterprotein, mKO (218 versus 236 amino acids), may experience stericinterferences that lead to poorer performance in similar fusions.The fused dimeric character of tdTomato effectively doubles its sizecompared to the monomeric orange fluorescent proteins, so sterichindrance is the most likely culprit in preventing tubulin localiza-tion. For most fusions, however, we observed little or no differencein performance between mOrange2 and mKO, suggesting thatmany proteins are more tolerant of fusion partners than tubulinor connexins.Though it already possessed reasonably good photostability,
TagRFP was still amenable to improvements by our photostabilityselection method. From a saturation-mutagenesis library of twochromophore-proximal residues (consisting of 400 independentclones), we selected a single clone with substantially enhancedphotostability. The selectedmutant, TagRFP-T, should prove to be avery useful addition to the fluorescent protein arsenal, as it is themost photostable monomeric fluorescent protein of any color yetdescribed under both arc-lamp and confocal laser illumination.As the applications of genetically encoded fluorescent markers
continue to diversify and become more complex, the demand forgreater photostability than is now available in fluorescent proteinshas likewise continued to grow. We expect our screening method to
be applicable to any of the existing fluorescent proteins and,with modifications, to be useful in selecting for more efficientphotoconvertible and photoswitchable fluorescent proteins aswell3,5,10,20,27–31. Possible enhancements to this selection techniquecould include time-lapse imaging of bacterial plates duringbleaching to enable direct selection for kinetics (independent ofabsolute brightness) and the use of higher-intensity illumi-nation from other light sources (such as lasers) duringscreening to select for or against nonlinear photobleachingbehavior. Ideally, a selection scheme that allows true simu-lation of microscopic imaging light intensities while maintaininga medium-to-high throughput should allow selection offluorescent proteins with the most beneficial properties forimaging applications.
METHODSMutagenesis. As the initial templates for library construction byrandom mutagenesis we used cDNA encoding mOrange2 andTagRFP (Evrogen)13, both of which had been previously humancodon–optimized. We performed error-prone PCR using theGeneMorph II kit (Stratagene) following the manufacturer’s pro-tocol, using primers containing BamHI and EcoRI sites formOrange variants or BamHI and BsrGI sites for TagRFP variants.We digested products of error-prone PCR products with appro-priate restriction enzymes and ligated the fragments into amodified pBAD vector (Invitrogen) or a constitutive bacterialexpression vector pNCS, both of which encode an N-terminal6His tag and linker identical to that found in pRSET B (Invitro-gen). We performed site-directed mutagenesis using the Quik-Change II kit (Stratagene) following the manufacturer’s protocolor by overlap-extension PCR. Sequences for all primers used inthis study are available in Supplementary Methods online.We transformed chemically competent or electrocompetentEscherichia coli strain LMG194 (Invitrogen) cells with librariesand grew them overnight at 37 1C on LB-agar supplemented with50 mg/ml ampicillin (Sigma) and 0.02% (wt/vol) L-arabinose(Fluka) (for pBAD-based libraries).
Library screening. For each round of random mutagenesis, wescreened 20,000–100,000 colonies (10–50 plates of bacteria), anumber sufficient to sample all possible single-site mutants and alimited number of double mutants. For each round of site-directedmutagenesis, we screened approximately threefold more coloniesthan the expected library diversity (for example, 1,200 colonies fora 400-member library) to ensure full coverage. We photobleachedwhole plates of bacteria for 10–120 min (determined empiricallyfor each round of directed evolution) on a Spectra-Physics92191–1000 solar simulator with a 1,600 W mercury arc lampequipped with two Spectra-Physics SP66239-3767 dichroic mir-rors to remove infrared and ultraviolet wavelengths. Remaininglight was filtered through 10-cm square bandpass filters (ChromaTechnology Corp.) appropriate to the fluorescent protein beingbleached (540/30 nm (B540/30; 525–555 nm) for mOrange- andTagRFP-based libraries or 568/40 nm (B568/40; 548–588 nm) formApple libraries). We measured final light intensities produced bythe solar simulator by a miniature integrating-sphere detector(SPD024 head and ILC1700 meter, International Light Corp.) tobe 95 mW/cm2 for the 540/30 filter and 141 mW/cm2 for the568/40 filter. We maintained the temperature of the bacterial plates
at 20 1C during solar simulator bleaching using a home-builtwater-cooled aluminum block. For mOrange mutant selection, weexamined the plates by eye as previously described32 using a150 W xenon lamp equipped with a 540/30 nm excitation filterand fiber optic light guides to illuminate the plates and 575 nmlong pass filter to visualize emission. For TagRFP mutant selection,we imaged the plates before and after bleaching on an imagingsystem (UVP) using 535/45 nm (512.5–557.5 nm) excitation and605/70 nm (570–640 nm) emission filters. In either case, we grewcolonies that maintained bright fluorescence after photobleachingand/or those with high post- to pre-bleach fluorescence ratios for8 h in 2 ml of LB medium supplemented with 100 mg/mlampicillin and then increased the culture volume to 4 ml withadditional LB supplemented with ampicillin and 0.2% (wt/vol)L-arabinose to induce fluorescent protein expression and grew thecultures overnight. We extracted protein from a fraction of eachcell pellet with B-PER II (Pierce) and obtained spectra using aSafire 96-well plate reader with monochromators (Tecan). Whenscreening for photostable variants, we obtained spectra before andafter photobleaching extracted protein on the solar simulator. Weextracted plasmid DNA from the remaining cell pellet with a mini-prep kit (Qiagen) and used it for sequencing.
Protein production and characterization. We expressed fluores-cent proteins from pBAD vectors in E. coli strain LMG194,purified them on Ni-NTA agarose (Qiagen) and characterizedthem as described2. Photobleaching measurements were per-formed on aqueous droplets of purified protein under oil asdescribed2,16. To determine whether the presence of molecularoxygen influenced bleaching, we performed our standard bleach-ing experiment before and after equilibrating the entire bleachingapparatus under humidified N2.
Additional methods. Primer list, descriptions of mass spectro-metry analysis, mammalian expression vectors, live-cell imagingand laser scanning confocal microscopy live-cell photobleachingare available in Supplementary Methods.
Accession numbers. GenBank: DQ336159 (mOrange2),DQ336160 (mApple) and EU582019 (TagRFP-T).
Note: Supplementary information is available on the Nature Methods website.
ACKNOWLEDGMENTSL.A. Gross performed mass spectroscopy. S.R. Adams performed gel filtrationexperiments. We thank R.E. Campbell and C.T. Dooley for helpful discussion.Sequencing services were provided by the University of California, San DiegoCancer Center shared sequencing resource and the Florida State UniversityBioanalytical and Molecular Cloning DNA Sequencing Laboratory. N.C.S. was aHoward Hughes Medical Institute predoctoral fellow during this work. This workwas additionally supported by the US National Institutes of Health (NS27177 andGM72033) and the Howard Hughes Medical Institute.
AUTHOR CONTRIBUTIONSN.C.S. designed the photostability selection protocol, performed all directedevolution and physical characterization of mApple and mOrange2, analyzed andorganized all data collected by other authors, and prepared the manuscript;M.Z.L. and M.R.M. performed directed evolution and physical characterizationof TagRFP-T; P.A.S. designed the home-built components of the solar simulatorapparatus and performed photobleaching measurements of purified proteins;K.L.H. and M.W.D. constructed mammalian expression vectors and performedall microscopy experiments involving live cells; R.Y.T. contributed to conceptualdevelopment, data analysis and manuscript preparation; all authors contributed toediting the manuscript.
COMPETING INTERESTS STATEMENTThe authors declare competing financial interests: details accompany the full-textHTML version of the paper at http://www.nature.com/naturemethods/.
Published online at http://www.nature.com/naturemethods/Reprints and permissions information is available online athttp://npg.nature.com/reprintsandpermissions
1. Campbell, R.E. et al. A monomeric red fluorescent protein. Proc. Natl. Acad. Sci.USA 99, 7877–7882 (2002).
2. Shaner, N.C. et al. Improved monomeric red, orange and yellow fluorescentproteins derived from Discosoma sp. red fluorescent protein. Nat. Biotechnol. 22,1567–1572 (2004).
3. Chudakov, D.M. et al. Photoswitchable cyan fluorescent protein for proteintracking. Nat. Biotechnol. 22, 1435–1439 (2004).
5. Habuchi, S. et al. Reversible single-molecule photoswitching in the GFP-likefluorescent protein Dronpa. Proc. Natl. Acad. Sci. USA 102, 9511–9516 (2005).
6. Karasawa, S., Araki, T., Nagai, T., Mizuno, H. & Miyawaki, A. Cyan-emitting andorange-emitting fluorescent proteins as a donor/acceptor pair for fluorescenceresonance energy transfer. Biochem. J. 381, 307–312 (2004).
7. Nagai, T. et al. A variant of yellow fluorescent protein with fast and efficientmaturation for cell-biological applications. Nat. Biotechnol. 20, 87–90 (2002).
9. Rizzo, M.A., Springer, G.H., Granada, B. & Piston, D.W. An improved cyanfluorescent protein variant useful for FRET. Nat. Biotechnol. 22, 445–449 (2004).
10. Wiedenmann, J. et al. EosFP, a fluorescent marker protein with UV-induciblegreen-to-red fluorescence conversion. Proc. Natl. Acad. Sci. USA 101,15905–15910 (2004).
11. Zapata-Hommer, O. & Griesbeck, O. Efficiently folding and circularly permutedvariants of the Sapphire mutant of GFP. BMC Biotechnol. 3, 5 (2003).
12. Ai, H.W., Henderson, J.N., Remington, S.J. & Campbell, R.E. Directed evolution ofa monomeric, bright and photostable version of Clavularia cyan fluorescentprotein: structural characterization and applications in fluorescence imaging.Biochem. J. 400, 531–540 (2006).
13. Merzlyak, E.M. et al. Bright monomeric red fluorescent protein with an extendedfluorescence lifetime. Nat. Methods 4, 555–557 (2007).
14. Matz, M.V. et al. Fluorescent proteins from nonbioluminescent Anthozoa species.Nat. Biotechnol. 17, 969–973 (1999).
15. Petersen, J. et al. The 2.0-A crystal structure of eqFP611, a far red fluorescentprotein from the sea anemone Entacmaea quadricolor. J. Biol. Chem. 278,44626–44631 (2003).
16. Shaner, N.C., Steinbach, P.A. & Tsien, R.Y. A guide to choosing fluorescentproteins. Nat. Methods 2, 905–909 (2005).
17. Miesenbock, G., De Angelis, D.A. & Rothman, J.E. Visualizing secretion andsynaptic transmission with pH-sensitive green fluorescent proteins. Nature 394,192–195 (1998).
18. Sinnecker, D., Voigt, P., Hellwig, N. & Schaefer, M. Reversible photobleaching ofenhanced green fluorescent proteins. Biochemistry 44, 7085–7094 (2005).
19. Habuchi, S. et al. Evidence for the isomerization and decarboxylation in thephotoconversion of the red fluorescent protein DsRed. J. Am. Chem. Soc. 127,8977–8984 (2005).
21. Andresen, M. et al. Structure and mechanism of the reversible photoswitch of afluorescent protein. Proc. Natl. Acad. Sci. USA 102, 13070–13074 (2005).
22. Andresen, M. et al. Structural basis for reversible photoswitching in Dronpa. Proc.Natl. Acad. Sci. USA 104, 13005–13009 (2007).
23. Henderson, J.N., Ai, H.W., Campbell, R.E. & Remington, S.J. Structural basis forreversible photobleaching of a green fluorescent protein homologue. Proc. Natl.Acad. Sci. USA 104, 6672–6677 (2007).
24. van Thor, J.J., Gensch, T., Hellingwerf, K.J. & Johnson, L.N. Phototransformationof green fluorescent protein with UV and visible light leads to decarboxylation ofglutamate 222. Nat. Struct. Biol. 9, 37–41 (2002).
25. Bell, A.F., Stoner-Ma, D., Wachter, R.M. & Tonge, P.J. Light-driven decarboxylationof wild-type green fluorescent protein. J. Am. Chem. Soc. 125, 6919–6926(2003).
26. van Thor, J.J., Georgiev, G.Y., Towrie, M. & Sage, J.T. Ultrafast and low barriermotions in the photoreactions of the green fluorescent protein. J. Biol. Chem.280, 33652–33659 (2005).
27. Verkhusha, V.V. & Sorkin, A. Conversion of the monomeric red fluorescent proteininto a photoactivatable probe. Chem. Biol. 12, 279–285 (2005).
28. Ando, R., Hama, H., Yamamoto-Hino, M., Mizuno, H. & Miyawaki, A. An opticalmarker based on the UV-induced green-to-red photoconversion of a fluorescentprotein. Proc. Natl. Acad. Sci. USA 99, 12651–12656 (2002).
30. Patterson, G.H. & Lippincott-Schwartz, J. Selective photolabeling of proteinsusing photoactivatable GFP. Methods 32, 445–450 (2004).
31. Tsutsui, H., Karasawa, S., Shimizu, H., Nukina, N. & Miyawaki, A. Semi-rationalengineering of a coral fluorescent protein into an efficient highlighter. EMBO Rep.6, 233–238 (2005).
32. Baird, G.S., Zacharias, D.A. & Tsien, R.Y. Circular permutation and receptorinsertion within green fluorescent proteins. Proc. Natl. Acad. Sci. USA 96,11241–11246 (1999).